Multi-LED Classification as Pretext For Robot Heading Estimation

TL;DR

This paper introduces a self-supervised method for robot detection and heading estimation using LED states, achieving competitive accuracy without extensive labeled data.

Contribution

It presents a novel self-supervised approach for robot heading estimation based on LED state classification, reducing reliance on labeled datasets.

Findings

Median image-space position error of 14 pixels

Relative heading MAE of 17 degrees

Supervised upperbound of 10 pixels and 8 degrees

Abstract

We propose a self-supervised approach for visual robot detection and heading estimation by learning to estimate the states (OFF or ON) of four independent robot-mounted LEDs. Experimental results show a median image-space position error of 14 px and relative heading MAE of 17 degrees, versus a supervised upperbound scoring 10 px and 8 degrees, respectively.